An anti-influenza combined therapy assessed by single cell RNA-sequencing

Abstract

Influenza makes millions of people ill every year, placing a large burden on the healthcare system and the economy. To develop a treatment against influenza, we combined virucidal sialylated cyclodextrins with interferon lambda and demonstrated, in human airway epithelia, that the two compounds inhibit the replication of a clinical H1N1 strain more efficiently when administered together rather than alone. We investigated the mechanism of action of the combined treatment by single cell RNA-sequencing analysis and found that both the single and combined treatments impair viral replication to different extents across distinct epithelial cell types. We showed that each cell type comprises multiple sub-types, whose proportions are altered by H1N1 infection, and assessed the ability of the treatments to restore them. To the best of our knowledge this is the first study investigating the effectiveness of an antiviral therapy against influenza virus by single cell transcriptomic studies.

Fig. 1. Combinatorial effect of 6’SLN-CD + IFN λ1 in HAE.

a Schematics of 6’SLN-CD and IFN λ1 (60 µg and 5.5 ng per tissue, respectively) combined administration. b Bar plot showing the kinetic of IAV replication in HAE treated with 6’SLN-CD only, or with IFN λ1 only, or with both compounds according to a. The results represent mean and standard deviation (calculated with the t-test analysis) of two independent experiments conducted in duplicate in HAE developed from a pool of donors and infected with 10³ RNA copies of clinical A/Switzerland/3076/2016 H1N1 (0 h corresponds to the time of viral inoculation). Viral replication was assessed measuring the apical release of IAV by RT-qPCR. **p ≤ 0.01; ***p ≤ 0.001; ****p ≤ 0.0001. HAE = human airway epithelia.

Fig. 2. Analysis of HAE cell diversity.

a t-distributed stochastic neighbor embedding (t-SNE) visualization of the major cell types composing human HAE. Individual cell types were annotated using a combination of graph-based clustering results from Seurat and expression analysis of several canonical cell-type-specific markers. The t-SNE plots shown in panels a–c are presented in the same spatial orientation (i.e., the location of cells expressing the canonical markers in panel b corresponds to the location of the specific cell types in panel a. b t-SNE plots illustrating in blue the expression patterns of some of the canonical markers used to annotate the three main airway epithelial cell types: FOXJ1 for ciliated cells, TCN1 for all secretory cells, TP63 for basal cells, LY6D for differentiating cells, SCGB1A1 for club cells and MUC5AC for goblet cells; scale bars are Log2. c t-SNE visualization of the scRNA-seq data for all single cells in the following conditions: mock steady sate, infected untreated (infected with A/Switzerland/3076/2016 H1N1 strain), infected + 6’SLN-CD, infected + IFN λ1, infected + IFN λ1 + 6’SLN-CD, and mock + IFN λ1 + 6’SLN-CD. d Heatmap representing the gene-expression profiles of 12,778 single cells from human HAE grouped into five clusters. The percentages of viral reads across cells are shown in red above the heatmap, while the number of total UMI counts is shown in light green. In each cluster, infected cells are ordered by increasing number of viral reads. Expression values are Pearson residuals from SCTransform binomial regression model [70] fitted to UMI counts (see Methods). The cells were clustered solely on the expression of the shown hallmark genes; HAE human airway epithelia, BdiS basal differentiating into secretory cells. e Violin plots showing the distributions of per-cell UMI (unique molecular identifiers) counts in HAE cell clusters. f Bar graph showing the relative percentage of each main epithelial cell type described above in each experimental condition described in c.

Fig. 3. Within-cell viral load across cell types and conditions (log odds scale).

a Violin plots of within-cell proportions of viral transcripts (VT) on log odds scale (by cell type and condition). Vertical dashed lines define different viral load classes: Z = zero VT; N = background noise of VT; L = low VT; M = medium VT; H = high VT (see Methods). b Fractions of cells in viral load classes within cell-type and condition groups. The thresholds defining the viral load classes as per (see methods) are shown as vertical dashed lines in panel a. c t-distributed stochastic neighbor embedding (t-SNE) visualization of fractions of cells in viral load categories, within cell-type and condition groups. Conditions as specified in Fig. 2c.

Fig. 4. Analysis of HAE cell types heterogeneity.

Stacked bar graph showing the relative percentages of HAE basal (a), b BdiS (basal differentiating into secretory), c secretory, and d ciliated subclusters across experimental conditions. Conditions as specified in Fig. 2c. HAE = human airway epithelia.